Classification based personality analysis on Turkish tweets


Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Faculty of Engineering, Department of Computer Engineering, Turkey

Approval Date: 2019

Student: Gökalp Maviş

Supervisor: İSMAİL HAKKI TOROSLU

Abstract:

Psychology researches suggest that some of the personality traits correlate with lin-guistic behavior and nowadays people most commonly present themselves to theworld by using social media. In social media, users openly reveal insights into theirlives and details about their personality. Although, there are many studies related tosocial media, only a small number of them work on personality prediction. In thisproject, we used not only text data but also the other statistics of users of Twitterwhich is one of the most commonly used social media platforms. The intent of thisthesis was modeling the correlation between Twitter data and Big Five PersonalityTraits by using machine learning algorithms. In this thesis, collected data, methodsof analyzing them and the machine learning techniques that successfully predict per-sonality are going to be described.